About Onfre
Support should be fast for customers and humane for the people doing it
Onfre exists because customers moved to WhatsApp and chat, and support teams were left answering 'where's my order?' at midnight. We think an AI agent should take that load, and do it honestly.
Why we're building it
Messaging changed what customers expect. They message a brand the way they message a friend, at any hour, and expect a reply in minutes. For a growing business, that means a support queue that never empties and a team that spends most of its day on the same handful of questions.
Early chatbots made this worse. They were built to deflect: rigid menus, scripted answers and no way out. Customers learned to type "agent" until something happened.
Large language models finally make a different kind of support agent possible, one that understands a messy, half-Hinglish question, finds the real answer in your own documents, takes the action, and recognises when a person should take over. That's what we're building.
Who it's for
Onfre is built for teams that do most of their support over chat: D2C brands, clinics, internet providers, education companies, travel and service businesses. Above all, it's built for teams that want to stay small without making customers wait.
Grounded or silent
Onfre answers from your sources or not at all. A confident wrong answer costs more trust than a slower right one.
A person is always one message away
No loops, no 'I didn't understand that' five times in a row. If a customer wants a human, they get one.
Your data stays yours
Your docs, tickets and customer conversations train your agent only, never a shared model and never anyone else's.
Measured by resolutions, not deflection
We bill for conversations that are actually solved and stay solved. Pushing customers away doesn't count.
Let Onfre take tonight's shift.
Connect WhatsApp, point Onfre at your help docs, and see it draft answers to your real tickets before any customer does.